🏠 About

This is the official implementation of the RAL paper Stable Trajectory Planning for Quadruped Robots using Terrain Features at Feet End.

s1_conpressed.mp4

s2.mp4

s3.mp4

real1.mp4

discrete_wooden_blocks.mp4

For the perceptive control architecture, please refer to legged_perceptive.

📚 Getting Started (simulation)

1. Install ocs2 following instructions from https://github.com/leggedrobotics/ocs2
2. Run the perceptive locomotion demo in ocs2.

   roslaunch ocs2_anymal_loopshaping_mpc perceptive_mpc.launch

4. Create your own terrain with grayscale. Assign the grayscale file path in the launch file.
5. Put the trajtories compouted with compute_body_traj.py into "ocs2/ocs2_robotic_examples/ocs2_perceptive_anymal/ocs2_anymal_loopshaping_mpc/data/".
6. Run the simulation.

   roslaunch ocs2_anymal_loopshaping_mpc perceptive_mpc.launch

🔗 Citation

If you find our work helpful, please cite:

@ARTICLE{li2026stable,
  author={Congfei Li and Shuyue Lin and Shenwei Qu and Zhuoyuan Liu and Qingjun Yang and Max Q.-H. Meng and Yuxiang Sun},
  journal={IEEE Robotics and Automation Letters}, 
  title={Stable Trajectory Planning for Quadruped Robots Using Terrain Features at Feet End}, 
  year={2026},
  volume={11},
  number={2},
  pages={2266-2273},
  doi={10.1109/LRA.2025.3645657}}

📄 License

This project is released under MIT license.

👏 Acknowledgements

• legged_perceptive